Higher Order Connection Enhanced Community Detection in Adversarial Multiview Networks

Area of Science:

• Network Science
• Data Mining
• Graph Theory

Background:

• Community detection in multiview networks is crucial but challenged by adversarial attacks.
• Existing methods struggle with hidden community structures due to deliberate edge removal.

Purpose of the Study:

• To develop a novel approach for robust community detection in adversarial multiview networks.
• To enhance intracommunity connections by leveraging higher-order network structures.

Main Methods:

• Proposed Higher Order Connection Enhanced Multiview Modularity (HCEMM).
• Discovered view-specific higher-order Microcommunities (VHM-communities).
• Added edges to VHM-communities to enhance intracommunity connectivity.

Main Results:

• HCEMM effectively discovers underlying community structures in multiview networks.
• The approach demonstrates robustness against adversarial attacks by recovering missing edges.
• Validated effectiveness across 16 real-world datasets.

Conclusions:

• HCEMM offers a significant advancement in community detection for adversarial multiview networks.
• Utilizing higher-order connections is key to overcoming adversarial manipulations.
• The method successfully recovers network structure and enhances detection accuracy.